Created
October 13, 2023 04:10
-
-
Save calebrob6/d32673997661a75d1a5e2e5c0afbe68a to your computer and use it in GitHub Desktop.
Short example of object detection training in TorchGeo.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| { | |
| "cells": [ | |
| { | |
| "cell_type": "code", | |
| "execution_count": 1, | |
| "id": "d7ff0785", | |
| "metadata": {}, | |
| "outputs": [], | |
| "source": [ | |
| "import torchgeo" | |
| ] | |
| }, | |
| { | |
| "cell_type": "code", | |
| "execution_count": 2, | |
| "id": "64d6ec01", | |
| "metadata": {}, | |
| "outputs": [ | |
| { | |
| "data": { | |
| "text/plain": [ | |
| "'0.6.0.dev0'" | |
| ] | |
| }, | |
| "execution_count": 2, | |
| "metadata": {}, | |
| "output_type": "execute_result" | |
| } | |
| ], | |
| "source": [ | |
| "torchgeo.__version__" | |
| ] | |
| }, | |
| { | |
| "cell_type": "code", | |
| "execution_count": 3, | |
| "id": "d91b04cd", | |
| "metadata": {}, | |
| "outputs": [], | |
| "source": [ | |
| "import torch\n", | |
| "from torchgeo.trainers import ObjectDetectionTask\n", | |
| "from torchgeo.datasets import VHR10\n", | |
| "from torch.utils.data import DataLoader\n", | |
| "import lightning.pytorch as pl\n", | |
| "import matplotlib.pyplot as plt" | |
| ] | |
| }, | |
| { | |
| "cell_type": "code", | |
| "execution_count": 4, | |
| "id": "8a565672", | |
| "metadata": {}, | |
| "outputs": [ | |
| { | |
| "name": "stdout", | |
| "output_type": "stream", | |
| "text": [ | |
| "Files already downloaded and verified\n", | |
| "loading annotations into memory...\n", | |
| "Done (t=0.02s)\n", | |
| "creating index...\n", | |
| "index created!\n" | |
| ] | |
| } | |
| ], | |
| "source": [ | |
| "def preprocess(sample):\n", | |
| " sample[\"image\"] = sample[\"image\"].float() / 255.0\n", | |
| " return sample\n", | |
| "\n", | |
| "ds = VHR10(root='data/VHR10/', split='positive', transforms=preprocess, download=True)" | |
| ] | |
| }, | |
| { | |
| "cell_type": "code", | |
| "execution_count": 5, | |
| "id": "4f4aab35", | |
| "metadata": {}, | |
| "outputs": [ | |
| { | |
| "data": { | |
| "text/plain": [ | |
| "650" | |
| ] | |
| }, | |
| "execution_count": 5, | |
| "metadata": {}, | |
| "output_type": "execute_result" | |
| } | |
| ], | |
| "source": [ | |
| "len(ds)" | |
| ] | |
| }, | |
| { | |
| "cell_type": "code", | |
| "execution_count": 6, | |
| "id": "abde065f", | |
| "metadata": {}, | |
| "outputs": [ | |
| { | |
| "data": { | |
| "text/plain": [ | |
| "{'image': tensor([[[0.3059, 0.3059, 0.3098, ..., 0.3765, 0.3686, 0.3647],\n", | |
| " [0.3020, 0.3020, 0.3098, ..., 0.3804, 0.3725, 0.3686],\n", | |
| " [0.2941, 0.2980, 0.3059, ..., 0.3804, 0.3686, 0.3569],\n", | |
| " ...,\n", | |
| " [0.4431, 0.4431, 0.4471, ..., 0.3373, 0.3333, 0.3333],\n", | |
| " [0.4431, 0.4431, 0.4471, ..., 0.3373, 0.3333, 0.3294],\n", | |
| " [0.4392, 0.4431, 0.4431, ..., 0.3412, 0.3333, 0.3294]],\n", | |
| " \n", | |
| " [[0.3490, 0.3490, 0.3529, ..., 0.4275, 0.4196, 0.4157],\n", | |
| " [0.3451, 0.3451, 0.3529, ..., 0.4314, 0.4235, 0.4196],\n", | |
| " [0.3333, 0.3373, 0.3451, ..., 0.4314, 0.4196, 0.4078],\n", | |
| " ...,\n", | |
| " [0.4941, 0.4941, 0.4980, ..., 0.3569, 0.3529, 0.3529],\n", | |
| " [0.4941, 0.4941, 0.4980, ..., 0.3569, 0.3529, 0.3490],\n", | |
| " [0.4902, 0.4941, 0.4941, ..., 0.3608, 0.3529, 0.3490]],\n", | |
| " \n", | |
| " [[0.1922, 0.1922, 0.1961, ..., 0.3098, 0.3098, 0.3059],\n", | |
| " [0.1882, 0.1882, 0.1961, ..., 0.3137, 0.3137, 0.3098],\n", | |
| " [0.1882, 0.1922, 0.2000, ..., 0.3137, 0.3098, 0.2980],\n", | |
| " ...,\n", | |
| " [0.4667, 0.4588, 0.4627, ..., 0.2314, 0.2275, 0.2275],\n", | |
| " [0.4667, 0.4588, 0.4627, ..., 0.2314, 0.2275, 0.2235],\n", | |
| " [0.4627, 0.4588, 0.4588, ..., 0.2353, 0.2275, 0.2235]]]),\n", | |
| " 'labels': tensor([1]),\n", | |
| " 'boxes': tensor([[563., 485., 629., 571.]]),\n", | |
| " 'masks': tensor([[[0, 0, 0, ..., 0, 0, 0],\n", | |
| " [0, 0, 0, ..., 0, 0, 0],\n", | |
| " [0, 0, 0, ..., 0, 0, 0],\n", | |
| " ...,\n", | |
| " [0, 0, 0, ..., 0, 0, 0],\n", | |
| " [0, 0, 0, ..., 0, 0, 0],\n", | |
| " [0, 0, 0, ..., 0, 0, 0]]], dtype=torch.uint8)}" | |
| ] | |
| }, | |
| "execution_count": 6, | |
| "metadata": {}, | |
| "output_type": "execute_result" | |
| } | |
| ], | |
| "source": [ | |
| "ds[0]" | |
| ] | |
| }, | |
| { | |
| "cell_type": "code", | |
| "execution_count": 7, | |
| "id": "47fe2f86", | |
| "metadata": {}, | |
| "outputs": [ | |
| { | |
| "data": { | |
| "text/plain": [ | |
| "(torch.Size([3, 808, 958]), torch.Size([3, 806, 950]))" | |
| ] | |
| }, | |
| "execution_count": 7, | |
| "metadata": {}, | |
| "output_type": "execute_result" | |
| } | |
| ], | |
| "source": [ | |
| "ds[0][\"image\"].shape, ds[1][\"image\"].shape" | |
| ] | |
| }, | |
| { | |
| "cell_type": "code", | |
| "execution_count": 8, | |
| "id": "38476acc", | |
| "metadata": {}, | |
| "outputs": [ | |
| { | |
| "data": { |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment